Aaron Lazor is the CEO of MyChargeBack.
It’s no secret—not among bankers, not among credit card network executives and, most certainly, not among cardholders. The chargeback process is broken, and that’s because it has not kept up with changes in the merchant environment.
The most significant change is the explosion in card-not-present (CNP) transactions due to the growing dominance of online commerce. E-commerce is inherently a generator of chargebacks because online purchasers cannot physically inspect their purchases beforehand. As a result, they may make mistakes in ordering or simply have buyer’s remorse afterward. Merchants, of course, can make mistakes of their own in shipping. In such an environment, chargebacks are a natural recourse.
However, mistakes are not the only factor that has spurred chargeback growth. The other one is the explosion in “friendly fraud.” Estimates vary, but friendly fraud is said to comprise anywhere between 40% and 80% of all chargeback requests. A December 2022 study by Sift found that 23% of all chargeback applicants admit to having engaged in friendly fraud. Moreover, a 2022 Merchant Fraud Journal report found that chargeback fraud is growing at a rate of 1.5% annually and noted that the cost of fighting it had already reached $25 billion in 2020.
Friendly fraud, of course, is not a mistake. It is a digital form of shoplifting. Consumers who knowingly purchase items (be it in person or online) that are exactly what they want and are exactly as advertised and then file a claim for a chargeback by claiming that the goods were not as ordered, arrived damaged or were never delivered at all are committing fraud. Their objective all along was to walk away with the item they purchased and with the money they paid for it back in their own wallets.
Needless to say, bank dispute departments—which are the fulcrum of the chargeback process—are overwhelmed. That is not solely because of the number of cases they have to handle. It is also because the chargeback process today is much the same as it was when it was originally designed almost a half-century ago. Unlike the rest of the banking ecosystem, it still remains manually intensive.
While technology is employed in the chargeback process, it is not done so in a consistent, end-to-end fashion. It does not connect all of the necessary data points or even assist in routine tasks. Most importantly, it is not used to assist the bank dispute department in making an informed analytic decision or to improve the customer experience.
Now that the banking industry is abuzz with talk of artificial intelligence (AI), the question naturally arises: Can AI fill the chargeback gap? The answer is that it could help, but it cannot fully solve the problem because AI, by definition, cannot replace people. It can only replace certain tasks that people have had to undertake until now.
There are plenty of tasks in the manual chargeback process that can be automated. However, perhaps 20% to 30% of all chargeback tasks cannot be automated since they require direct communication with cardholders and merchants. The most obvious is the decisioning of the case itself since the cardholder’s arguments have to be compared to those of the merchant in the context of the chargeback guidelines established by the card network, which are frequently updated.
However, there is plenty of room for AI to substantially replace much of the 70% to 80% that is left.
One obvious example would be auto-checking the validity of the documentation that has to be submitted in advance for a chargeback to proceed. That includes verifying that the credit card in question has not expired, whether the transaction dates are within the permitted time limits, whether the merchant did indeed charge and receive the sums, whether the merchant already gave a refund to the same card number, and whether an unusual number of similar disputes have been raised with the merchant—which would be prima facie evidence that this is not an example of friendly fraud.
Another critical function that AI can address is in aiding consumers to avoid chargebacks. At present, banks are ill-informed about warnings that financial regulatory agencies issue about fake investment opportunities, unlicensed brokerages and the like. As a result, they don’t generally act to interdict credit card transactions between consumers and the subjects of the regulator warnings.
Using AI to integrate these regulatory warnings into banks’ processing systems would allow them to trigger a pop-up warning to the cardholder before the transaction is completed. This would substantially reduce not only consumer fraud but also the number of chargebacks currently straining the resources of bank dispute departments.
AI could also make a major impact when it comes to cybersecurity and sanctions enforcement. AI could integrate information from a wide variety of sources that would provide a reasonable determination that the merchant is legitimate. A merchant that claims to have a London address but whose revenue is transferred to a bank in Estonia and whose DNS records indicate its IP numbers are Russian could automatically be tagged for further investigation before the transaction is approved.
Nonetheless, any viable AI model that could be incorporated into the chargeback process would have to integrate a trigger mechanism that could allow the cardholder to directly engage with a bank dispute department representative when necessary. At the very least, the introduction of AI might just be able to reduce the number of superfluous cases that are clogging up the system and compromising the quality of service that cardholders with legitimate disputes deserve.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Read the full article here
